Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Genralized loading to multiple data loaders #340

Open
wants to merge 3 commits into
base: OpenKE-PyTorch
Choose a base branch
from
Open

Genralized loading to multiple data loaders #340

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
1531d91
Update Base.cpp
howardchanth Sep 29, 2021
c85f33c
Update Reader.h
howardchanth Sep 29, 2021
1e979f0
Update TrainDataLoader.py
howardchanth Sep 29, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 21 additions & 1 deletion openke/base/Base.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -168,12 +168,32 @@ void sampling(
INT negRate = 1,
INT negRelRate = 0,
INT mode = 0,
INT domain_idx=0,
bool filter_flag = true,
bool p = false,
bool val_loss = false
) {
pthread_t *pt = (pthread_t *)malloc(workThreads * sizeof(pthread_t));
Parameter *para = (Parameter *)malloc(workThreads * sizeof(Parameter));

// Update the global variables to current loader index
trainList = trainLists[domain_idx];
trainTotal = trainTotals[domain_idx];

trainHead = trainHeads[domain_idx];
trainTail = trainTails[domain_idx];
trainRel = trainRels[domain_idx];

lefHead = lefHeads[domain_idx];
rigHead = rigHeads[domain_idx];
lefTail = lefTails[domain_idx];
rigTail = rigTails[domain_idx];
lefRel = lefRels[domain_idx];
rigRel = rigRels[domain_idx];

entityTotal = entityTotals[domain_idx];
relationTotal = relationTotals[domain_idx];

for (INT threads = 0; threads < workThreads; threads++) {
para[threads].id = threads;
para[threads].batch_h = batch_h;
Expand All @@ -199,4 +219,4 @@ void sampling(
int main() {
importTrainFiles();
return 0;
}
}
42 changes: 42 additions & 0 deletions openke/base/Reader.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,6 +6,7 @@
#include <algorithm>
#include <iostream>
#include <cmath>
#include <vector>

INT *freqRel, *freqEnt;
INT *lefHead, *rigHead;
Expand All @@ -14,6 +15,25 @@ INT *lefRel, *rigRel;
REAL *left_mean, *right_mean;
REAL *prob;

// Vector of collecting the global variables from different data loaders
std::vector< Triple* > trainLists = std::vector<Triple*>();

std::vector< Triple* > trainHeads = std::vector<Triple*>();
std::vector< Triple* > trainTails = std::vector<Triple*>();
std::vector< Triple* > trainRels = std::vector<Triple*>();

std::vector< INT* > lefHeads = std::vector<INT*>();
std::vector< INT* > rigHeads = std::vector<INT*>();
std::vector< INT* > lefTails = std::vector<INT*>();
std::vector< INT* > rigTails = std::vector<INT*>();
std::vector< INT* > lefRels = std::vector<INT*>();
std::vector< INT* > rigRels = std::vector<INT*>();

std::vector< INT > trainTotals = std::vector<INT>();
std::vector< INT > tripleTotals = std::vector<INT>();
std::vector< INT > entityTotals = std::vector<INT>();
std::vector< INT > relationTotals = std::vector<INT>();

Triple *trainList;
Triple *trainHead;
Triple *trainTail;
Expand Down Expand Up @@ -83,6 +103,16 @@ void importTrainFiles() {
trainRel = (Triple *)calloc(trainTotal, sizeof(Triple));
freqRel = (INT *)calloc(relationTotal, sizeof(INT));
freqEnt = (INT *)calloc(entityTotal, sizeof(INT));

// Collect the address of train lists, thier heads, tails and relations
trainLists.push_back(trainList);
trainHeads.push_back(trainHead);
trainTails.push_back(trainTail);
trainRels.push_back(trainRel);

// Collect number of entities and relations of this data loader
entityTotals.push_back(entityTotal);
relationTotals.push_back(relationTotal);
for (INT i = 0; i < trainTotal; i++) {
tmp = fscanf(fin, "%ld", &trainList[i].h);
tmp = fscanf(fin, "%ld", &trainList[i].t);
Expand Down Expand Up @@ -117,6 +147,18 @@ void importTrainFiles() {
rigRel = (INT *)calloc(entityTotal, sizeof(INT));
memset(rigHead, -1, sizeof(INT)*entityTotal);
memset(rigTail, -1, sizeof(INT)*entityTotal);

// Collect left and rights of heads, tail and relations
lefHeads.push_back(lefHead);
rigHeads.push_back(rigHead);
lefTails.push_back(lefTail);
rigTails.push_back(rigTail);
lefRels.push_back(lefRel);
rigRels.push_back(rigRel);

// Collect train totals
trainTotals.push_back(trainTotal);

memset(rigRel, -1, sizeof(INT)*entityTotal);
for (INT i = 1; i < trainTotal; i++) {
if (trainTail[i].t != trainTail[i - 1].t) {
Expand Down
7 changes: 6 additions & 1 deletion openke/data/TrainDataLoader.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,6 +52,7 @@ def __init__(self,
ctypes.c_int64,
ctypes.c_int64,
ctypes.c_int64,
ctypes.c_int64,
ctypes.c_int64
]
self.in_path = in_path
Expand All @@ -71,6 +72,7 @@ def __init__(self,
self.negative_ent = neg_ent
self.negative_rel = neg_rel
self.sampling_mode = sampling_mode
self.domain_idx = domain_idx
self.cross_sampling_flag = 0
self.read()

Expand Down Expand Up @@ -115,6 +117,7 @@ def sampling(self):
self.negative_ent,
self.negative_rel,
0,
self.domain_idx,
self.filter,
0,
0
Expand All @@ -137,6 +140,7 @@ def sampling_head(self):
self.negative_ent,
self.negative_rel,
-1,
self.domain_idx,
self.filter,
0,
0
Expand All @@ -159,6 +163,7 @@ def sampling_tail(self):
self.negative_ent,
self.negative_rel,
1,
self.domain_idx,
self.filter,
0,
0
Expand Down Expand Up @@ -226,4 +231,4 @@ def __iter__(self):
return TrainDataSampler(self.nbatches, self.cross_sampling)

def __len__(self):
return self.nbatches
return self.nbatches